blis

Yes, if you have two AVX512 FMA units, you use them for GEMM; with a current microarchitecture implementation, if you only have one, you don't. https://github.com/flame/blis/blob/2d8ec164e7ae4f0c461c27309... GEMM is one of relatively few computations with sufficient computational intensity.

Why this obsession with MKL on AMD hardware? I've long made measurements of OpenBLAS on "large" dimension serial DGEMM, at least. Even on Intel hardware between Westmere and SKX, with the exception of KNL, it was always at least within the sort of noise level of HPC jobs of MKL performance, and was always better than ACML on the generations of Opterons we used. There are results for AVX2 systems with older versions of OpenBLAS and BLIS, which is AMD's BLAS at https://github.com/flame/blis/blob/master/docs/Performance.m... I haven't seen or made measurements for EPYC2 yet (and don't have results with the current versions online for Haswell and SKX) but I'd be surprised if AMD BLIS doesn't perform equally well on EPYC2. OpenBLAS serial DGEMM is very similar to MKL on SKX, for instance, with BLIS not far behind. I think AMD work has con

You can see BLIS on Intel's home turf at https://github.com/flame/blis/blob/master/docs/Performance.m... (52-core SKX) and compare with OpenBLAS on 32-core Zen1. (Multithreaded BLAS isn't typically used in HPC, where the parallelism is elsewhere.)

This is an overstatement. ICC consistently compiles the slowest and produces the largest binaries. It also defaults to something close to -ffast-math, which may or may not be appropriate. If your app benefits from aggressive inlining and vectorization at the expense of potentially huge increases in code size, ICC is likely to do well for you. However, I've seen lots of cases where well-vectorized code is faster with GCC or Clang, including some very important cases using Intel intrinsics. (Several such cases reported to/acknowledged by Intel; some have been fixed over the years, but these observations are not uncommon.) BLIS is used by AMD and is a good open alternative to MKL (for BLAS) across many platforms. https://github.com/flame/blis/blob/master/docs/Performance.m...

Netlib BLAS is a very low bar [1], and not at all how one should go about writing a performance portable BLAS. BLIS (https://github.com/flame/blis/) is a much better approach, and underlies vendor implementations on AMD (https://developer.amd.com/amd-aocl/blas-library/) and many embedded systems. [1] GEMV is entirely limited by memory bandwidth, thus quite uninteresting from a vectorization standpoint. Maybe you meant GEMM?